From sthibault at debian.org  Mon Jul 20 10:33:12 2026
From: sthibault at debian.org (Samuel Thibault)
Date: Mon, 20 Jul 2026 11:33:12 +0200
Subject: [Tts-project] Migration from Berkeley DB
In-Reply-To: <875x2blg0h.fsf@poretsky.localdomain>
References: <ahMnKT3IRTvJig-1@end> <87jyrdfrjy.fsf@poretsky.localdomain>
 <ak7IBmaxkamA2rSe@end> <87ldbelphh.fsf@poretsky.localdomain>
 <algB5vMHXB3UoamW@end> <87ech2l0yf.fsf@poretsky.localdomain>
 <alpscyQdQF6f_Tql@end> <87a4rokzl7.fsf@poretsky.localdomain>
 <alvRGMK28DLSn5Y1@end> <875x2blg0h.fsf@poretsky.localdomain>
Message-ID: <al3rWJyyt9WqJEW1@end>

Hello,

Igor B. Poretsky, le dim. 19 juil. 2026 08:12:14 +0300, a ecrit:
> >>>>> "Samuel" == Samuel Thibault <sthibault at debian.org> writes:
> 
>     Samuel> That is it. But that's a common thing to do. And notably on
>     Samuel> architectures which can natively run both 32b and 64b
>     Samuel> variants. Setting multi-arch tells people they can really
>     Samuel> run both versions concurrently.
> 
> Thus, the most simple and natural thing to do from my point of view is
> to drop out the Multi-Arch field particularly for the rulex-data
> package, leaving it untouched for all other binary packages built from
> the rulex source.

That should be fine enough, yes.

The non-native librulexdb2 variants won't work, but at least they will
be usable for cross-builds.

> But is it correct to have Multi-Arch field not for all binary packages
> built from one source?

Yes.

Samuel


From poretsky at mlbox.ru  Mon Jul 20 20:08:59 2026
From: poretsky at mlbox.ru (Igor B. Poretsky)
Date: Mon, 20 Jul 2026 22:08:59 +0300
Subject: [Tts-project] Migration from Berkeley DB
In-Reply-To: <al3rWJyyt9WqJEW1@end> (Samuel Thibault's message of "Mon, 20 Jul
 2026 11:33:12 +0200")
References: <ahMnKT3IRTvJig-1@end> <87jyrdfrjy.fsf@poretsky.localdomain>
 <ak7IBmaxkamA2rSe@end> <87ldbelphh.fsf@poretsky.localdomain>
 <algB5vMHXB3UoamW@end> <87ech2l0yf.fsf@poretsky.localdomain>
 <alpscyQdQF6f_Tql@end> <87a4rokzl7.fsf@poretsky.localdomain>
 <alvRGMK28DLSn5Y1@end> <875x2blg0h.fsf@poretsky.localdomain>
 <al3rWJyyt9WqJEW1@end>
Message-ID: <87fr1dscl0.fsf@mlbox.ru>

Hello,

>>>>> "Samuel" == Samuel Thibault <sthibault at debian.org> writes:

    >> Thus, the most simple and natural thing to do from my point of
    >> view is to drop out the Multi-Arch field particularly for the
    >> rulex-data package, leaving it untouched for all other binary
    >> packages built from the rulex source.

    Samuel> That should be fine enough, yes.

Done. And, btw, the other my packages (freespeech, ru-tts and
multispeech) are also fully updated and synchronized with it.

    Samuel> The non-native librulexdb2 variants won't work, but at least
    Samuel> they will be usable for cross-builds.

Speaking exactly, they will even be pretty functional, but not with the
data from the rulex-data package installed for other
architecture. Indeed, nothing in the librulexdb2 package obligates it to
use just the data from the rulex-data package. Though yes, the main
purpose of Multi-Arch of the rulexdb2 package is cross-building.

Best regards,
Igor.